AI companion robot Abi, piloted in an Australian senior care home in 2025, is now coming to US facilities amid a crowded field of rivals and a $173,000 institutional robot called Moya — raising both promising acceptance rates and sharp ethical questions about cost, data, and human connection.
Northwestern University engineers printed flexible artificial neurons from graphene and molybdenum disulfide that successfully triggered responses in living mouse brain tissue, an early step toward energy-efficient neuromorphic computing and future brain-machine interfaces.
Deep Origin says its AI platform, which blends physics-based simulation with machine learning, found viable cancer immunotherapy drug candidates at a 31% hit rate, nearly 100 times the industry's typical rate, though no AI-designed drug has yet won regulatory approval.
Genetic testing giant Natera has launched a proprietary AI foundation model platform built on more than 250,000 tumor exomes and over a million plasma timepoints, aiming to predict which cancer patients will respond to immunotherapy before treatment starts.
Cedars-Sinai has embedded OpenEvidence's AI clinical reference platform directly inside Epic, giving physicians, nurses, and pharmacists patient-context-aware answers grounded in medical literature at the point of care.
PathAI received FDA Breakthrough Device Designation for PathAssist Derm, an AI tool aimed at dermatopathology, a specialty where diagnostic agreement between pathologists falls below 50% on the hardest skin lesion cases.
British startup C the Signs is seeking first-of-its-kind FDA De Novo clearance for an AI tool that mines existing medical records to flag cancer risk, after helping identify 75,000 cancer patients across the UK's NHS.
Artera received FDA clearance for ArteraAI Breast, a digital pathology AI tool that scores recurrence risk for HR+/HER2-negative breast cancer patients using existing biopsy images rather than a separate genomic test.
Ibex Medical Analytics won its first FDA clearance for Prostate Detect, an AI pathology tool whose validation data showed it caught prostate cancer in 13% of cases pathologists had initially missed.
Wood Mackenzie projects utilities will build only about 28% of the 1,066 gigawatts requested for AI data centers, with the rest amounting to duplicative 'phantom load' that could distort grid planning.